les-mis-nlp-search

PUT les-miserable-embedded
{
  "mappings": {
    "properties": {
      "ml": {
        "properties": {
          "inference": {
            "properties": {
              "model_id": {
                "type": "keyword"
              },
              "predicted_value": {
                "type": "dense_vector",
                "dims": 384,
                "index": true,
                "similarity": "l2_norm"
              }
            }
          }
        }
      }
    }
  }
}

# Useful setup if required
PUT _ingest/pipeline/embed-les-mis
{
  "description" : "embed pipeline",
    "processors" : [
      {
        "inference" : {
          "model_id" : "sentence-transformers__msmarco-minilm-l-12-v3",
          "field_map": {
            "paragraph": "text_field"
          }
        }
      },
      {
        "set" : {
          "field" : "timestamp",
          "value" : "{{_ingest.timestamp}}"
        }
      }
    ]
}

POST _reindex
{
  "source": {
    "index": "les-mis"
  },
  "dest": {
    "index": "les-miserable-embedded",
    "pipeline": "embed-les-mis"
  }
}

POST _ml/trained_models/sentence-transformers__msmarco-minilm-l-12-v3/deployment/_infer
{
  "docs": [{"text_field": "What was Jean Valjean prisoner number?"}]
}

GET les-miserable-embedded/_search?size=1
GET les-miserable-embedded

POST les-miserable-embedded/_knn_search
{
  "_source": ["paragraph", "line"],
  "knn": {
    "field": "ml.inference.predicted_value",
    "k": 5,
    "num_candidates": 10,
    "query_vector": # Put query vector here
  }
}

POST _ml/trained_models/sentence-transformers__msmarco-minilm-l-12-v3/deployment/_infer
{
  "docs": [{"text_field": "How much could Jean Valjean sell the candlesticks?"}]
}